This panel took place at February's iMedia Brand Summit in Florida. Rick Parkhill, President of iMedia Communications, kicked it off:
Rick Parkhill: We now have a panel focused on market mix modeling. These are the decisions that are determining how marketers are going to allocate their funds over the next 10 years. Doug Weaver, President, Upstream Group, will introduce this panel. Take it away, Doug.
Doug Weaver: Thank you Rick. Welcome. Welcome. So we’re going to spend about 40 minutes together here and hopefully we’ll get quite a bit of interaction with you guys. I know there’s going to be a lot of questions on this topic because it’s something that’s going to determine a very important outcome for all of us, and that’s how the money gets distributed over the course of the next 10 or 15 years
This morning in Joe Uva’s remarks, he really painted what I thought was a brilliant picture of this brave new universe of this totally integrated media world. It’s no surprise to anybody here that there is just an onslaught of new devices, new channels, new points of distribution. And it is a mess to try and figure out how to spend money in that world.
Yesterday on the panel that Neil Perry so eloquently moderated, you heard Stephen Blumberg and Erik Whiteford talk about media mix modeling briefly. And they talked about how they use modeling to really predict outcomes, and essentially hedge their bets on where they were going to spend money in the future for their brands. Marketing mix modeling is what we’re here to talk about today. We’re here to talk about: where interactive fits within that marketing mix modeling world; how it can be done well; is it relevant; and, is it something that marketers are going to be able to use; and, how should they be using it, in order to make the right decisions going forward.
Now, having said that, we know what you’re thinking because this topic certainly could be a little bit of a challenge. I’ll make one disclaimer here before we get started. Okay? I’m joined up here by two of the most brilliant minds in the research space. And do you remember that song -- who watched "Sesame Street" as a kid? Give me some applause. (Applause) Okay. You remember that song, "One of these things is…"?
Audience: "…not like the others."
Weaver: Right. One of these things is not like the other things. So, I’m going to be here as your Sherpa and hopefully we’re going to be able to translate what’s going on up here for everyone here. But I think the gentlemen who have joined me, who I’ll introduce in a moment, are just fabulous at putting this stuff into layman’s terms and translating some really brilliant research concepts.
I’m going to actually turn around here and introduce the two gentlemen who are joining me. I’m not going to read the bios, but I’ll talk a little bit about what they do and how long they’ve been doing it.
On my immediate left is Gerard Broussard from MOne. He is the Senior Partner and Director of Media Analytics for MOne. This is an agency that has put its money where its mouth is. Gerard runs the largest agency group of its kind in the business. And he has been just a force in our business. Formerly with Ogilvy, he has been at this for how many years?
Gerard Broussard: Total?
Weaver: Total.
Broussard: Do you mean online and offline -- everything?
Weaver: Everything.
Broussard: Twenty-four years.
Weaver: Twenty-four years of media research and modeling experience. He also chairs MOne’s new Media Committee. He joined MOne in 1998 after several successful years at Ogilvy. And we’re very, very pleased to have him.
On my far left here is John Nardone, a familiar face in a new role. A lot of you may know John from his days at Modem Media. Back when I was selling some of the first ads on the web in 1994, John was the guy who was pitching those to his clients at Modem, and was a real force there. He has since moved on and joined Marketing Management Analytics, MMA, which is a leading vendor in this space, and a firm that is really pioneering a new chapter in the world of marketing mix modeling, which he’s going to talk a little bit about today. He has been, again, somebody that’s really been committed to the interactive side of the business and now really committed to the entire interactive integrated marketing modeling space.
So, please welcome both John and Gerard. (Applause)
When we were talking about this session on the phone, one of the things I asked John and Gerard to do is, I said we really need to reach out and make sure that everybody in the audience is really connected with this topic. So I thought, as a proxy for the person in the audience who might not be as up on marketing mix modeling as he might want to be, I’m using my Aunt Grace as a proxy. I’m going to start off with John and I’m going to ask him to explain to Aunt Grace what this is about. Tell me what marketing mix modeling is and give us some perspective on it.
John Nardone: Okay. Basically marketing mix modeling is a mathematical technique that’s used to quantify the impact in the marketplace of not just media factors, but all your marketing levers. Essentially it looks at pricing, external factors like seasonality, the impact of your PR, your television, your internet -- and how those things all work together to actually drive sales for your business.
There’s a huge range of data that can be considered and pulled into a marketing mix model. From, obviously, the media factors, direct mail, email, press and events, to qualitative factors like brand awareness, and operational factors like distribution, quality. And even external factors like the state of the economy, competitive media, seasonality. Anything that we can get a data stream from to represent, we can pull into the model and essentially use the data to drive the analytics behind the methodology. And what we hopefully get from TV GRPs, radio GRPs, in-store promotions, we distill out: "What are the factors that are responsible for driving the sales over time?" And we wind up looking at this on a weekly basis, if at all possible, and relating those changes in sales to changes in those causal factors. This is all based on statistics and calculus. And the really neat thing about it is that it takes into effect some of the effects of advertising over time. So if you think about advertising when GRPs started to hit the marketplace, it takes a while for consumer awareness to build. And when that TV campaign goes off air, the effect of that advertising doesn't just stop; it remains in people's minds and continues to influence the market over time. And so you have a build and a decay curve that occurs. And this methodology helps to understand those build and decay curves and saturation points, and gives us a lot of tools for managing how much we spend and where we spend it.
Weaver: Okay. Not bad. Not bad. But I think Aunt Grace still wants to know more. Gerard can you flesh that out a little bit more for us?
Broussard: My turn to answer to Aunt Grace. Okay. I think from a user standpoint, what do we use market mix models for? Basically it’s not only just about media placement, but as John mentioned before, it’s about total marketing budget allocations and all the things that -- and market factors -- that go into let’s say, influencing a sale. And it could help you determine the split, let’s say, between marketing, PR and promotion, outside of the advertising world.
Then there’s media channel allocation, which we're really focusing on in this particular workshop. And the question becomes: How much influence does TV, print and the internet have on sales -- or the combination of media? We look for media combinations and the interaction effects when we're modeling the data over time. Then overlaid on top of all of that we can put costs -- the cost to buy a TV ad, the total budget of TV advertising versus internet versus print -- and we come up with efficiency measures that help determine the return on investment for each media.
And then finally there’s the question of timing. I’ll have an example in a second, but when are the best times to advertise? Probably when the product is in greatest demand.
Weaver: If I can just pipe in -- so it's not just a set of tools to determine media allocation dollars, but also to inform the decisions about how and when to spend that money and so forth?
Broussard: And even how to price the product, too.
Weaver: Okay.
Broussard: Has anybody ever seen a trial like this before? I’m probably not going to get too many hands, but … Let me describe it very simply. This is an example of how sales might lay out for a soft drink over a two-year period. And the two mountains that you see basically, the peaks are the summer parts of the year. And then the valleys are when it gets cold. Obviously soft drink sales go up when the weather gets better. And then, what you see there is a little tiny red line that goes over the peaks and valleys -- that’s your sales. The bars are actually describing what is influencing those sales. In this case, the big drivers are weather, the price of the product -- of the soft drink -- and then the TV advertising. And you can see the TV advertising represented as the red bar. That’s driving a lot of the sales there. And then yellow as weather, the sun. And then the green bars as being price. In the following year, TV does not show up as much. It could have been a change in copy or maybe they pulled the TV ads away. And something like outdoor might have more of an influence, or newspapers.
So what we try to do is we try to decompose the effects of media and the marketplace over a long period of time, and explain what actually drove sales. This is one way to do it.
Tomorrow: How do you compare the effect of interactive if it's getting less money to begin with?
